athena-code
Health Uyari
- License — License: MIT
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 6 GitHub stars
Code Uyari
- network request — Outbound network request in overlay/packages/opencode/src/installation/athena-release.ts
Permissions Gecti
- Permissions — No dangerous permissions requested
Bu listing icin henuz AI raporu yok.
Open-source terminal AI coding agent with persistent local memory, automatic context recall, and searchable cross-session history.
Athena Code
Athena Code is an open-source terminal AI coding agent with persistent memory,
automatic context recall, and searchable history across coding sessions.
Athena Code is a memory-focused fork of
OpenCode. It gives an AI coding agent
local, durable memory; freezes relevant context for a session; recalls related
facts for each turn; and indexes prior conversations for later search.
Project status: early development. Native release builds are produced for
Linux, macOS, and Windows on x64 and ARM64. macOS and Windows binaries are not
yet code-signed.
Why Athena Code?
Most coding agents start each session without durable knowledge of your
preferences, architecture decisions, or previous work. Athena Code adds a
local-first memory layer directly to the agent loop:
- Persistent AI memory: save stable facts, preferences, decisions, and
workflow notes for future sessions. - Automatic contextual recall: retrieve relevant memories for the current
request without sending the entire memory store on every turn. - Cross-session, cross-agent search: index and search previous
conversations with SQLite FTS5 — Athena Code's own, plus local Claude Code,
Codex, opencode, and Hermes session stores when present. - Frozen session context: build one deterministic memory snapshot per
session and reuse it across tool calls. - Native agent tools: expose
memory_write,memory_read, andsession_recallto the model. - Local-first storage: memory and recall data stay in local files by
default; the memory hot path requires no Athena backend service. - Open agent loop: built on the open-source OpenCode terminal coding agent.
Quick Start
Linux and macOS
Install the latest build for the current operating system and architecture:
curl -fsSL https://raw.githubusercontent.com/luckeyfaraday/athena-code/main/scripts/install.sh | bash
The installer verifies the release checksum, copies the executable to~/.local/share/athena-code/bin/athena-code, and creates~/.local/bin/athena-code. The installed command does not depend on a cloned
repository.
Install a specific version:
curl -fsSL https://raw.githubusercontent.com/luckeyfaraday/athena-code/main/scripts/install.sh |
bash -s -- --version v0.2.1
See GitHub Releases
for available versions, archives, checksums, and release notes.
Windows
Run in PowerShell:
irm https://raw.githubusercontent.com/luckeyfaraday/athena-code/main/scripts/install.ps1 | iex
The installer verifies the release checksum, installsathena-code.exe under %LOCALAPPDATA%\AthenaCode, and adds its bin
directory to your user PATH.
Updating
Installed builds update themselves. On startup Athena Code checks the latest
GitHub release in the background and shows an update prompt inside the app;
accepting it downloads, verifies, and installs the new version (restart to run
it). You can also update from the command line at any time:
athena-code upgrade # latest release
athena-code upgrade v0.2.1 # specific version
Set "autoupdate": false in the global config to disable the startup check,
or "autoupdate": "notify" to be notified without ever auto-installing.
Build from source
Requirements:
- A supported Linux, macOS, or Windows host
- Git
- Node.js with
npx - At least 5 GB of free temporary disk space
- Visual Studio 2022 Build Tools with Desktop development with C++ on
Windows
git clone https://github.com/luckeyfaraday/athena-code.git
cd athena-code
./scripts/build.sh
The build checks out the pinned OpenCode revision, applies Athena Code's patch
and source overlay, installs dependencies, and writes the executable toruntime-bin/<platform>-<architecture>/.
Install a local Linux or macOS build by passing its generated path:
./scripts/install.sh --from-file ./runtime-bin/linux-x64/athena-code
On Windows:
.\scripts\install.ps1 -FromFile .\runtime-bin\windows-x64\athena-code.exe
Usage
Start the interactive terminal UI in the current project:
athena-code
Run a one-shot coding task:
athena-code run "explain this repository and identify the highest-risk module"
Save and inspect durable memories:
athena-code memory add "Use pnpm for this organization"
athena-code memory list
Resume a session:
athena-code --session SESSION_ID
Use athena-code --help for the complete command reference.
How Memory Works
Athena Code separates memory by lifecycle so relevant context is available
without rebuilding or resending everything continuously.
| Layer | Purpose | Lifecycle |
|---|---|---|
| Global memory | Stable user facts, preferences, and decisions | Persists across projects and sessions |
| Project memory | Repository-specific facts and context | Stored inside the project |
| Frozen snapshot | Bounded memory context for one agent session | Built once and reused byte-for-byte |
| Turn recall | Memories relevant to the current request | Recomputed once per user turn |
| Session index | Searchable excerpts from previous conversations across local agents (Athena Code, Claude Code, Codex, opencode, Hermes) | Indexed locally with SQLite FTS5, rescanned incrementally in the background |
Memory locations:
- Global memory:
~/.athena-code/memory/entries.jsonl - Project memory:
<project>/.context-workspace/memory/entries.jsonl - Global cross-project session search index:
~/.athena-code/context/sessions.db
Set ATHENA_CODE_HOME to change the global Athena Code data directory.
Athena Code tells the model to treat recalled text as background data rather
than as newer instructions. Do not store passwords, API keys, private keys, or
other secrets in agent memory.
For the design rationale, see
Athena Turn-Ownership Memory Design.
Architecture
This repository maintains a reproducible fork rather than vendoring the entire
upstream source tree:
OpenCode pinned revision
+
patches/opencode-athena.patch
+
overlay/packages/opencode/...
=
runtime-bin/<platform>-<architecture>/athena-code
Key directories:
| Path | Contents |
|---|---|
overlay/ |
Athena-owned memory, recall, tools, and TUI source |
patches/ |
Integration and branding changes applied to OpenCode |
scripts/build.sh |
Reproducible source build |
scripts/install.sh |
Linux and macOS installer |
scripts/install.ps1 |
Windows PowerShell installer |
test/ |
Memory, recall, snapshot, and session-index tests |
docs/ |
Technical design documentation |
Development
Run the focused test suite:
npx --yes [email protected] test test
Build the complete executable:
./scripts/build.sh
See CONTRIBUTING.md for the development workflow and pull
request expectations.
Platform Support
| Platform | Status |
|---|---|
| Linux x64 | Native release build |
| Linux ARM64 | Native release build |
| macOS Apple Silicon | Native release build; unsigned |
| macOS Intel | Native release build; unsigned |
| Windows x64 | Native release build; unsigned |
| Windows ARM64 | Native release build; unsigned |
Platform support describes Athena Code packaging, not every platform supported
by upstream OpenCode.
Unsigned macOS builds may require confirmation in System Settings > Privacy &
Security on first launch. Windows SmartScreen may also warn about an
unrecognized publisher. Code signing and macOS notarization are planned release
hardening work.
FAQ
Is Athena Code the same as OpenCode?
No. Athena Code is an independent, memory-focused fork built from a pinned
OpenCode revision. It retains OpenCode's terminal coding-agent foundation and
adds Athena-owned persistent memory, context snapshots, recall, session search,
tools, branding, tests, and release packaging.
Does Athena Code require the Athena desktop application?
No. Athena Code builds and runs as a standalone terminal application. It can
also be launched by Athena or another terminal workspace manager.
Is memory uploaded to a server?
Athena Code's native memory store, snapshots, and session index are local by
default. Model requests still go to whichever AI provider you configure through
the underlying OpenCode provider system.
Can Athena Code remember information across repositories?
Yes. Global memories under ~/.athena-code are available across folders.
Project memories and indexed session history remain scoped to their project.
How is this different from adding instructions to AGENTS.md?
AGENTS.md is best for repository instructions that should always be loaded.
Athena Code memory is designed for durable facts and selective retrieval:
relevant entries are recalled for a request while unrelated entries remain out
of the prompt.
Project Policies
License and Attribution
Athena Code is available under the MIT License. It is derived from
OpenCode, which is also MIT licensed.
See THIRD_PARTY_NOTICES.md for upstream attribution.
Athena Code is an independent project and is not affiliated with or endorsed by
the OpenCode maintainers.
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi